Unverified50% confidenceFactExact time
2016年击败李世石的AlphaGo Lee使用了约16万局人类职业棋手对弈数据进行监督学习预训练
1
Sources
50%
Confidence
Long-term
Relevance
8/7/2026
First Seen
Sources
nanoAlphaZero：单文件JAX实现24小时训练2700分棋力
Related Claims
UnverifiedThe 2019 AlphaStar faced imperfect information, real-time decision-making, and an action space of up to 10^26, and its league training mechanism made the total training volume equivalent to 200 years of human gameplay experience.59% similarUnverifiedAlphaStar采用多智能体自我博弈和联盟训练机制，其训练量相当于人类玩家200年的游戏经验58% similarUnverifiedAlphaGo's defeat of the human Go champion is a landmark application of reinforcement learning.58% similarUnverifiedAlphaStar在2019年在《星际争霸II》欧服达到大师级水平（前0.15%），通过模仿学习与多智能体强化学习的结合实现56% similarUnverifiedAfter automatic optimization by Claude, most models participating in the comparison were able to complete LoRA training within 16 GB of VRAM.55% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/700727API
curl https://kongchang.com/api/v1/knowledge/claims/700727MCP
get_claim(id=700727)